Better locations.
Sharper unit economics.

Challenge

Placement was guesswork. Venues that looked right on a walk-through often underperformed once installed, and there was no way to rank corridors or vet a location before the sales visit — so every pitch was a gamble on foot traffic, install feasibility and owner appetite.

Approach

A two-stage engine. First, score the Kiez: feeder-street density, business momentum and rent trajectory rank every neighbourhood corridor. Then drill to venue level — footfall, installability and revenue gap per address. Where open data had no usable address, we reconstructed one from the road network and neighbouring businesses, lifting coverage from 70% to ~99%.

Models

12,000+ businesses across Berlin and Brandenburg classified into 13 sectors and 89 categories, with every street scored for advertising value and rankings anchored on measured pedestrian counts rather than assumptions. The Prenzlauer Berg pilot put the Schönhauser Allee corridor first at 77.8/100 — feeder-street density 100/100, ~19,000 pedestrians a day.

Built to scale

The engine is city-agnostic by design: it runs on open street, business and footfall data, so a new market is a data load rather than a new project. The same 13-sector / 89-category taxonomy re-scores any venue type — cafés, gyms, retail franchises, hotels, medical practices — and the two-stage logic (rank the corridor, then the address) holds whether the target is 20 screens in one district or a national rollout. Address reconstruction is what makes it portable: coverage stays near-complete even where open data is patchy.

Challenge

A growing operation with no spatial visibility into where loyal versus churning customers clustered, and no data-driven way to choose new service areas for its restaurant network.

Approach

We standardized internal order data alongside open demographic and competitor datasets, then ran spatial RFM analyses on order frequency, basket size and total spend to segment customers by location.

Models

Two parallel growth scenarios — one weighted toward residents, one toward daytime workers — were correlated to surface priority expansion zones and forecast service-area profitability for restaurants in the network.

Challenge

Urban datasets related to planning, infrastructure, demographics, transportation and land use existed across disconnected systems and in multiple formats, limiting the municipality's ability to compare, analyze and manage information collectively. The absence of an integrated spatial data infrastructure made it difficult to produce coordinated, evidence-based planning and service decisions at the neighborhood and district scales.

Approach

Datasets were collected from municipal departments, public institutions and open-data sources through institutional coordination and official data requests. Raster, CAD, textual and GIS datasets were standardized, georeferenced and spatialized within a unified coordinate system.

Models

The resulting 50 GB geodatabase was organized under 15 main thematic categories — administrative boundaries, socio-economic structure, election data, land use, transportation, infrastructure, geological structure, natural assets, conservation areas, zoning plans and urban transformation — with 400+ subcategories, layers and thematic datasets enabling integrated visualization, querying, classification and spatial analysis.

Challenge

Pedestrianization decisions within the Historic Peninsula were often evaluated through fragmented approaches focused primarily on physical street conditions. The district's complex urban structure, tourism intensity, historical layers and pedestrian dynamics required an integrated methodology capable of identifying streets with the highest pedestrian-oriented transformation potential.

Approach

Multiple spatial datasets were produced, standardized and integrated into a road-network database. Variables such as street slope, cultural inventory density, non-residential land-use intensity, public transportation mobility and survey-based mobility perception were weighted to generate a pedestrianization / walkability score for each street segment.

Models

Using spatial interpolation, network analysis and multi-criteria evaluation (AHP), all datasets were analyzed through street centerline geometries — enabling a comparative assessment of pedestrian-oriented transformation potential across the Historic Peninsula.